MABLE: a framework for learning from natural instruction

ABSTRACT

The Modular Architecture for Bootstrapped Learning Experiments (MABLE) is a system that is being developed to allow humans to teach computers in the most natural manner possible: by using combinations of descriptions, demonstrations, and feedback. MABLE is a highly modular, well-engineered, and extendable system that provides generalized services, such as control, knowledge representation, and execution management. MABLE works by accepting instruction from a teacher and forms concrete learning tasks that are fed to state-of-the-art machine learning algorithms. To make the learning tractable, specialized heuristics, in the form of learning strategies, are used to derive bias from the instruction. The output of the learning is then incorporated into the system's background knowledge to be used in performing tasks or as the basis for simplifying the process of learning difficult concepts.

Although still in development, MABLE has already demonstrated the ability to learn four different types of knowledge (definitions, rules, functions, and procedures) from three different modes of student/teacher interaction on two separate, qualitatively different domains. MABLE presents a unique opportunity for machine learning researchers to easily plug in and test algorithms in the context of instructible computing. In the near future, MABLE will be freely available as an open source project.

REFERENCES

• 1: J. F. Allen, N. Chambers, G. Ferguson, L. Galescu, H. Jung, M. D. Swift, and W. Taysom. Plow: A collaborative task learning agent. In AAAI, pages 1514--1519. AAAI Press, 2007.
• 2: Jaime Carbonell , Oren Etzioni , Yolanda Gil , Robert Joseph , Craig Knoblock , Steve Minton , Manuela Veloso, PRODIGY: an integrated architecture for planning and learning, ACM SIGART Bulletin, v.2 n.4, p.51-55, Aug. 1991 [doi>10.1145/122344.122353]
• 3: D. Corkill. Blackboard Systems. AI Expert, 6(9), January 1991.
• 4: Daniel Oblinger. Bootstrapped Learning -- External Materials. http://www.sainc.com/bl-extmat/, October 2008.
• 5: S. B. Huffman and J. E. Laird. Flexibly instructable agents. Journal of Artificial Intelligence Research, 3:271--324, 1995.
• 6: Hiroaki Kitano , Minoru Asada , Yasuo Kuniyoshi , Itsuki Noda , Eiichi Osawa, RoboCup: The Robot World Cup Initiative, Proceedings of the first international conference on Autonomous agents, p.340-347, February 05-08, 1997, Marina del Rey, California, United States [doi>10.1145/267658.267738]
• 7: John E. Laird , Allen Newell , Paul S. Rosenbloom, SOAR: an architecture for general intelligence, Artificial Intelligence, v.33 n.1, p.1-64, Sept. 1987 [doi>10.1016/0004-3702(87)90050-6]
• 8: Douglas B. Lenat , R. V. Guha , Karen Pittman , Dexter Pratt , Mary Shepherd, Cyc: toward programs with common sense, Communications of the ACM, v.33 n.8, p.30-49, Aug. 1990 [doi>10.1145/79173.79176]
• 9: David Morley , Karen Myers, The SPARK Agent Framework, Proceedings of the Third International Joint Conference on Autonomous Agents and Multiagent Systems, p.714-721, July 19-23, 2004, New York, New York [doi>10.1109/AAMAS.2004.267]
• 10: Open Source Initiative. Open Source Initiative OSI - The BSD License:Licensing. http://www.opensource.org/licenses/bsd-license.php, October 2008.
• 11: SpringSource. Spring Framework. http://springframework.org/, October 2008.
• 12: Gerald Jay Sussman, A Computer Model of Skill Acquisition, Elsevier Science Inc., New York, NY, 1975

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.6 Learning

General Terms:
Design, Experimentation

Keywords:
MABLE, architecture, learning